Tokens por archivo generado ≈ 2 500
Archivos/mes                     = 10 000
Consumo total                    = 25 M tokens
Costo tokens                     = 25 M / 1 000 × 0,0007 = $17,50
GPU A100 2 h/mes (0,02 USD/h)    = $0,04
Total mensual ≈ $22‑$25

# 1️⃣ Instala el cliente de MAI‑Code
pip install azure-ai-ml

# 2️⃣ Configura la conexión (usa Managed Identity)
az login --identity
export AZURE_SUBSCRIPTION_ID=$(az account show --query id -o tsv)

# 3️⃣ Llama al modelo
python - <<'PY'
from azure.ai.ml import MLClient
from azure.identity import DefaultAzureCredential

ml_client = MLClient(
    credential=DefaultAzureCredential(),
    subscription_id="${AZURE_SUBSCRIPTION_ID}",
    resource_group_name="rg-iai",
    workspace_name="ws-iai"
)

prompt = """
Crea un endpoint FastAPI llamado /items que reciba un JSON con los campos:
- name (str)
- price (float)
- tags (list[str], opcional)
Incluye modelo Pydantic y manejo de errores 400.
"""
response = ml_client.online_endpoints.invoke(
    endpoint_name="mai-code",
    request_body={"input": prompt, "max_tokens": 500}
)
print(response)
PY

from fastapi import FastAPI, HTTPException
from pydantic import BaseModel, Field
from typing import List, Optional

app = FastAPI()

class Item(BaseModel):
    name: str = Field(..., example="Laptop")
    price: float = Field(..., gt=0, example=999.99)
    tags: Optional[List[str]] = None

@app.post("/items")
async def create_item(item: Item):
    if item.price <= 0:
        raise HTTPException(status_code=400, detail="Price must be positive")
    return {"message": "Item created", "item": item.dict()}

# Prompt sencillo
prompt="Refactoriza este bucle for en JavaScript usando .map():\n\nlet result = [];\nfor (let i = 0; i < arr.length; i++) { result.push(arr[i] * 2); }"

const result = arr.map(x => x * 2);

name: Generar código con MAI‑Code

on:
  push:
    paths:
      - 'prompts/**'

jobs:
  mai-code:
    runs-on: ubuntu-latest
    permissions:
      id-token: write   # necesario para Managed Identity
    steps:
      - uses: actions/checkout@v3

      - name: Login en Azure con OIDC
        uses: azure/login@v1
        with:
          client-id: ${{ secrets.AZURE_CLIENT_ID }}
          tenant-id: ${{ secrets.AZURE_TENANT_ID }}
          subscription-id: ${{ secrets.AZURE_SUBSCRIPTION_ID }}

      - name: Instalar cliente MAI‑Code
        run: pip install azure-ai-ml

      - name: Generar código
        env:
          PROMPT_FILE: prompts/${{ github.sha }}.txt
        run: |
          python generate.py $PROMPT_FILE generated/${{ github.sha }}.py

      - name: Commit y push
        uses: stefanzweifel/git-auto-commit-action@v4
        with:
          commit_message: "🤖 Código generado por MAI‑Code"
          branch: main